On the reliability of Microsoft Excel XP for statistical purposes

The paper deals with the reliability of Microsoft Excel XP for statistical purposes. In section 1 we show that Excel can compute negative variances. In section 2 we see that the random numbers generated by Excel cannot suffice scientific statistical requirements, and in section 3 we have to report that unbearable errors with the computation of statistical distributions that have been reported to Microsoft and later on have been published (Knusel, 1998) are still to be found in Microsoft Excel XP. So one has again to warn the scientific community against using Microsoft Excel for statistical purposes.